BR102015015579A2 - method and device for determining a set of modifiable elements in an image group - Google Patents

Abstract

Method and Device for Determining a Set of Modifiable Elements in an Image Group A method for determining a set of modifiable elements in an image group of an encoded bit stream representative of an audio and video content is disclosed. the method comprises determining (s12) a set of candidate modifiable elements wherein a candidate modifiable element comprises a modified encoded bit stream value and a spatial spread map associated with the modified value, a spatial spread map (sp6) comprising pixels whose decoding is impacted when the modified value is placed in the encoded bitstream; determining (s14) a heatmap for each reference frame, the heatmap comprising, for each reference frame pixel, information representative of the use of said pixel for temporal prediction during decoding of said portion of the encoded bit stream. encoding a group of images; obtain (s16) a set of modifiable elements between candidate modifiable elements, a modifiable element having a spatial spread map that does not overlap the corresponding heat map.

Description

“METHOD AND AVAILABLE FOR DETERMINING A SET OF MODIFIED ELEMENTS IN A PICTURE GROUP” TECHNICAL FIELD

In the following, a method for determining a set of modifiable elements in the video watermark signaling domain is disclosed. Specifically, a method for determining a set of modifiable elements in a group of images of an encoded bit stream representative of audio and video content is disclosed, wherein the method comprises determining a heat map. The corresponding device is also revealed.

BACKGROUND

[002] This section is intended to introduce the ieitor into various aspects of the art which may relate to various aspects of the present principles which are described and / or claimed below. It is believed that this discussion is useful in providing the reader. with prior information to facilitate a better understanding of the various aspects of these principles. Therefore, it should be understood that these statements are to be read with this understanding, not as prior art admissions.

Traitor tracking consists of serving clients with watermarked content with a unique identifier. If a pirated copy is later discovered on an unauthorized distribution network, then it is possible to identify the misbehaving client.

Among known methods for marking watermarked content, one approach comprises directly modifying the bit stream in compressed format without any decompression. A key challenge then is to ensure that the changes introduced do not significantly impair video quality at the decoding stage. The introduction of advanced entropy coding standards, such as context adaptation and arithmetic coding, for example, has taken such more challenging bitstream modifications. A single bit stream modification can result in catastrophic degradation if no proper care, especially regarding spatial and temporal propagation, is taken. When a frame, used for prediction, is watermarked, the changes introduced may be propagated temporarily through the decoding process, and may thus compromise watermark imperceptibility because of uncontrolled propagation of the watermark. 'water introduced.

Temporal propagation, also referred to as temporal displacement, originates from temporal prediction, a mechanism routinely used in video codecs. It can affect a large number of pixels in a large number of frames: (i) in a given frame, multiple pixels may use the same reference pixel in their predictions, and (ii) different frame pixels may also use the same pixel. of reference. In addition, subpixel interpolations involved in temporal predictions further amplify the effect of modification.

[G06] Bitstream analysis to identify the set (position, value) where watermarks can be inserted into the inlay stage is also known as watermark preprocessing.

[007] A first watermark preprocessing solution is available with bitstream video watermarked signaling systems that watermark non-reference coded frames (such as non-reference B-frames). codecs), thus avoiding any temporal propagation of changes introduced by construction. However, frames of reference are also of great interest. In fact, being able to watermark reference frames would increase the inlay rate and thus the watermark robustness. In addition, many distribution networks deal with streams that use reference frames only to have an alignment between decoding and rendering frame orders that imply more flexible streaming capabilities. Therefore, watermarking these frames of reference is of great interest to bitstream watermarked systems, but it certainly increases the challenges with respect to temporal propagation management.

For reference frames, such as frames used to predict other frames (frame t, frames 8, or reference frames P), a second solution, as disclosed in US20130188712, is to compute compression influence values that indicate, for each macroblock, how much it influences the pixel values of other macroblocks through intraframe or interframe prediction. However, constructing these compression influence values is very challenging in practice and very computationally intensive. This actually requires tracking the influence of a block on several frames in time. Macroblocks with small influence values are then selected for watermarked signaling. That being said, US2013Q1S8712 does not disclose any mechanism to prevent the temporal propagation of different changes from overlapping, possibly resulting in uncontrollable visual artifacts.

A variant of this solution is disclosed in FR2949283A1. An image including a macroblock not serving as a reference for decoding the next and previous images of the video stream (i.e. during intraframe or interframe prediction) is dynamically identified between the video images in the video stream. This non-reference macroblock is then replaced by an alternate macroblock encoding the desired identifying information, i.e. the watermark F ·: r ίο, the FR2949283A1 failure c does not reveal 'explr * p.' "Ur * and how the macroblocks not serving as a reference are identified and thus does not solve the challenge of computing the compression influence values.

An efficient computation method for determining candidate watermarks that manages the temporal and spatial propagation of watermarks because of prediction is necessary for this reason. For example, we might consider watermarks (position, value) whose spatial propagation does not contain any macroblock used for prediction by another frame. This is a slightly tighter selection than the ideal solution, also known as selecting a set of candidate watermarks whose spatiotemporal propagation maps do not intersect, but it can be computed somewhat efficiently.

SUMMARY

[011] The present principles overcome at least one of the disadvantages of the prior art by proposing a method for determining a modifiable set of elements of an encoded bit stream representative of audio and video content wherein the method comprises determining a heat map. representative of the use of each pixel as a reference.

In fact, a notable idea of the present principles is to watermark regions in reference frames without any or with controlled temporal propagation comprising in a first step either analyzing the bit stream to find out all the locations eligible to signal along with the associated surrogate characteristics and values (spatial propagation, and some robustness and fidelity criteria) for each frame capable of being watermarked when constructing, for each reference frame, a map representing the use of each pixel as a reference, that is, a pixel involved in some predictions from another frame, called a heatmap or reference heatmap. Then, in a second step, the set of correct locations is selected as the one that meets the robustness and fidelity criteria and whose spatial propagation maps are restricted within non-reference regions of the reference frames given by the corresponding reference heat maps. .

For this purpose, a method for determining a set of modifiable elements in a group of images of an encoded bit stream representative of audio and video content is disclosed. Those skilled in the art will appreciate that a modifiable element does not correspond to a syntax element, but to a bit stream segment encoding the content. In fact, with CABAC entropy encoding, a syntax element is encoded in a noninteger number of bits. As a result, the bit stream segment associated with a modifiable element can contribute to or even encode a single syntax element, but rarely encode a single syntax element entirely. In other words, a modifiable element comprises an alternative value of fixed length (for example, a binary 2-byte extension word) at a given position in the bit stream. The method comprises determining a set of candidate modifiable elements comprising a position, a modified bit stream value encoded at this position, and a spatial propagation map associated with the modified value; determining a heat map comprising, for each pixel of a frame of reference, information representative of the use of the pixel for temporal prediction during decoding of the encoded bit stream; obtaining a modifiable element between candidate modifiable elements of the encoded bit stream. according to the heat map and spatial propagation map of the modifiable element.

[014] In one embodiment, information representative of the use of a pixel for temporal prediction is a binary value indicating whether or not the pixel is used for temporal prediction. In another embodiment, information representative of the use of a pixel for temporal prediction comprises the number of times the pixel is used for temporal prediction. Also in another variant, information representative of the use of a pixel for temporal prediction comprises the sum of the prediction weights associated with the pixels using the prediction reference pixel. Each variant is advantageously adapted to one variant with respect to propagation management. watermarks as described hereinafter.

Advantageously, a heat map is determined for reference frames where candidate modifiable elements are present. In an advantageous embodiment, such reference frames are limited to P frames and B reference frames.

According to a specific embodiment, determining a heat map comprises, for each decoded frame of the image group: initializing information representative of the use of a pixel for temporal prediction, for example, as zero; decode said frame; and for each decoded macroblock of said decoded frame, updating the information representative of the use of a pixel for temporal prediction in the heat map (s) associated with the reference frame (s) that the macroblock decoded point.

According to another specific embodiment, obtaining a modifiable element comprises, for each frame of the group of scanned images in reverse order, selecting a candidate modifiable element whose spatial propagation map does not overlap with the temporal heat map of the frame; and updating information representative of the use of a pixel for temporal prediction for all pixels used as macroblock reference of the spatial spread map associated with the modified value of the selected candidate modifiable element. Advantageously, this second step to obtain modifiable element is performed in reverse decoding order. For each frame, the step selects the set of candidate modifiable elements that both meet the robustness and fidelity criteria and propagate to unreferenced reference frame regions, as given by the corresponding reference heat maps; and then updates the previous reference color maps according to the pixels used to predict the macroblocks of the selected elements.

[018] According to a first variant, selecting an element modifies ve! The candidate comprises discarding candidate modifiable elements when the heat map value of at least one pixel of the spatial propagation map is not equal to zero. Advantageously, this variant only maintains watermarks that do not induce any temporal propagation. According to a second variant, selecting a candidate modifiable element comprises, for each candidate modifiable element, determining a first dependency value comprising the number of pixels in the spatial propagation map of the candidate modifiable element that are used for prediction according to the map. of heat; and discard all candidate modifiable elements having their first dependency value greater than a first threshold. Advantageously, this variant retains elements that are likely to induce limited temporal propagation since only a controlled number of spatial propagation map pixels are used for prediction by other frames. According to a third variant, selecting a candidate modifiable element comprises, for each candidate modifiable element, determining a second dependency value comprising the sum of information representative of the use of a temporal prediction pixel associated with all pixels in the spatial propagation map. of the candidate modifiable element; and discard candidate modifiable elements whose second dependency value exceeds a second threshold. Advantageously, this variant selects elements that produce controlled temporal propagation. For example, this variant ensures that a watermark will not directly affect more pixels in other frames than a specified value. λ ». A device for determining a set of modifiable elements in a group of images of an encoded bit stream representative of audiovisual content is disclosed. The device comprises at least one processor configured to determine a set of candidate modifiable elements comprising a position. in the encoded bit stream, a modified value and a spatial propagation map associated with the modified value; for determining a heat map comprising, for each pixel of a frame, information representative of the use of the pixel for temporal prediction during decoding of the encoded bit stream; and for selecting a modifiable element among the candidate modifiable elements of the bitmap encoded according to the heatmap and the modifiable element's spatial propagation map, [020] A device for determining a set of modifiable elements in an image group of an encoded bit stream representative of audiovisual content is disclosed. The device comprises features for determining a set of candidate modifiable elements comprising a position in the encoded bit stream, a modified value and a spatial propagation map associated with the modified value; features for determining a heat map comprising, for each pixel of a frame, information representative of the use of the pixel for temporal prediction during decoding of the encoded bit stream; and features for selecting a modifiable element among the candidate modifiable elements of the coded bit stream according to the heat map and the spatial propagation map of the modifiable element.

A computer program product comprising program code instructions for performing the processing method steps, according to any of the embodiments and variants disclosed, when this program is run on a computer. j022] A processor readable media having instructions stored therein to induce a processor to perform at least the steps of the processing method according to any of the disclosed embodiments and variants. Although not explicitly described, the present embodiments may be employed in any combination or subcombination.For example, any variant of the information carried by the heatmap may be combined with the selection step or the heatmap update step. In addition, any feature or variant described for the method is compatible with a device intended for processing the disclosed methods and a computer readable storage media for storing program instructions.

BRIEF DESCRIPTION OF DRAWINGS

Other features and advantages of the present principles will appear with the description of a non-limiting embodiment of the present principles, which will be illustrated with the help of the accompanying drawings: Figure 1 represents a device for determining a set of modifiable elements of according to a specific and non-limiting embodiment of the invention; Figure 2 represents an exemplary architecture of the device of Figure 1, according to a specific and non-limiting embodiment of the present principles; Figure 3 is a flow chart of a method for determining a set of modifiable elements according to a specific and non-limiting embodiment of the present principles; Figure 4 is a detail of the flow chart shown in Figure 3, according to a specific and non-limiting embodiment of the present principles; Figure 5 is a detail of the flow chart shown in Figure 3, according to a specific and non-limiting embodiment of the present principles; Figure 6 illustrates the determination of the chaot maps according to a specific non-limiting embodiment of the present principles; and Figure 7 illustrates the selection of a modifiable element according to a spatial propagation map and a heat map according to a specific and non-limiting embodiment of the present principles.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS Figure 1 is a device 1 for determining a set of elements modifiable according to a specific and non-limiting embodiment of the present principles. Device 1 comprises an input 10 configured to receive at least one group of images made from a number of I, P and B frames. Images I, P, B may be obtained from a source. According to different embodiments of the present principles, the source belongs to a set comprising: - a local memory, for example, a video memory, a RAM, a flash memory, a hard disk; - a storage interface, for example a mass storage interface, a ROM, an optical disc or a magnetic medium; - a communication interface, for example, a wired interface (for example, a bus interface, an extended area network interface, a local area network interface) or a wireless interface (such as an IEEE 802.11 interface). or a Bluetooth interface); and an image capture circuit (e.g., a sensor such as, for example, a CCD (or Charge Coupled Device) or CMOS (or Complementary Metal Oxide Semiconductor)).

Input 10 is connected to a module 12 configured to determine a set of candidate modifiable elements or candidate watermarks. Each candidate modifiable element comprises * a position in the encoded bit stream, a modified value, and a spatial propagation map associated with this modified value. The spatial propagation map is described hereinafter. Input 10 is also connected to a module 14 configured to determine heat maps. Advantageously the module 14 comprises a decoder. Module 12 determining candidate modifiable elements and module 14 determining heat maps operate in parallel or in any order. Module 16 is configured to obtain modifiable elements based on spatial propagation maps and heat maps. Module 16 is connected to an output 18. Modifiable elements can be stored in memory or can be sent to a watermark embedding. As an example, modifiable elements are stored in remote memory or in a location, for example, video memory or RAM, a hard disk. In one embodiment, the modifiable elements are sent to a watermark embed via a storage interface, for example a mass storage interface, a ROM, a fiash memory, an optical disk, or a magnetic storage. and / or transmitted via a communication interface, for example, an interface to a peer-to-peer link, a communication bus, a point-to-multipoint link, or a broadcast network.

[Q26] Figure 2 represents an exemplary architecture of device 1 according to a specific and non-limiting embodiment of the present principles. Processing device 1 comprises one or more processors 110, which one (s) are, for example, a CPU, GPU and / or Digital Signal Processor (DSP) along with internal memory 120 (for example, RAM, ROM, EPROM). Processing device 1 comprises one or more input / output interfaces 130 adapted to display output information and / or allow a user to enter commands and / or data (e.g., a keyboard, a mouse, a touch surface, a network camera, a dial); and a power source 140 that may be external to the processing device 1. The processing device 1 may also comprise network interface (s) (not shown). According to an exemplary and non-limiting embodiment of the present principles, processing device 1 further comprises a computer program stored in memory 120. The computer program comprises instructions which, when executed by processing device 1, in particular processor 110 , cause the processing device 1 to perform the processing method described in Figure 3. In one embodiment, the computer program is stored externally to the processing device 1 on a non-transient digital data carrier, for example on a media. external storage such as an HDD, CD-ROM, DVD, a read-only and / or DVD drive, and / or a DVD read-write drive, all known in the art. The processing device 1 thus comprises an interface for reading the computer program. Additionally, device 1 can access one or more Universal Serial Bus (USB) type storage devices (for example, "memory cards '') via corresponding USB ports (not shown).

According to exemplary and non-limiting embodiments, processing device 1 is a device belonging to a set comprising: a mobile device; - a communication device; - a gaming device; - a tabtet (or laptop); - a laptop; - a photographic camera; - a video camera; - an encoding chip; - a decoding chip; - a photo server; - a video server (for example, a broadcast server, an on-demand video server or a network server); and - a video sharing platform.

FIG. 3 is a flowchart of a method for determining a set of modifiable elements according to a specific and non-limiting embodiment of the present principles.

[028] In a step S12, a set of candidate modifiable elements is obtained, for example, by module 12 implementing a watermarked signaling algorithm for each bit stream frame encoding a group of images. A candidate modifiable element comprises a position in the encoded bit stream, where watermarks will be inserted later in the inlay stage, along with an associated alternative value. Hereafter, the pair (position, alternative value) is referred to as the modified value and both terminologies are used interchangeably. A candidate modifiable element also comprises the induced spatial propagation map. In one embodiment, a candidate modifiable element further comprises features such as some robustness and fidelity criteria. Advantageously, any prior art watermarked signaling algorithm is compatible with the present principles for determining a candidate modifiable element. Such an algorithm, for example, is described in WO 2010/021682 A1. Those skilled in the art will appreciate that in the variant with respect to the H264 watermark signaling algorithm, where the mark is carried by a motion vector in the bit stream, the spatial propagation associated with an alternative value of a motion vector results from the prediction of an alternate motion vector block. As shown in Fig. 7, SP6 spatial propagation is activated by macroblock 700 of table i ~ 6. whose motion vector may be changed to an alternative value by the watermarked signaling process and thereby may cause changes in neighboring blocks. The SP6 spatial propagation map comprises all pixels whose decoding is impacted by using the alternative motion vector value for 700, that is. the pixels of macroblock 700 and its neighboring right and bottom macroblock in this particular case. In another variant of MPEG2, those skilled in the art will appreciate that the propagation map is insignificant and corresponds to the alternative value itself. Frames where tags are searched are called frames that can be marked with watermarks. In one embodiment, frames I are not frames capable of being watermarked. A bit stream decoding encoding the image group is thus required in this step.

[029] In a step S14, a heat map is determined. The heat map comprises, for each pixel of a reference frame, information representative of the use of the pixel for temporal prediction during decoding of the encoded bit stream. Advantageously, heat maps are not constructed for each frame of the image group, but only for frames where candidate modifiable elements are present and which are used as a reference. Thus, in a variant where modified values correspond to motion vectors, heat maps are not constructed for I frames since they do not host motion vectors. In a variant where modified values correspond to DCT coefficients, it is pertinent to construct heat maps for frames I. Heat maps need not be constructed for non-reference frames although these frames include candidate modifiable elements. However, for practical reasons, it may be convenient to also set heat maps for non-reference frames, which are empty by default. Advantageously, heat maps are constructed for P-frames and B-reference frames, which can then be introduced into the GOP subset of squares where candidate modifiable elements can be discovered. Such heat maps associate a value with each pixel of the corresponding frame and therefore have the same dimension (width and height) as the video frames. Advantageously, heat maps are stored as value tables wherein the indices in the table correspond to the position of the pixel in the frame. Cafor map values are derived by collecting information throughout the decoding of GOP macroblocks. In a first variant, the camphor map comprises, for each pixel of the associated reference frame, information representative of the use of this pixel for temporal prediction, that is, one or more macroblocks in other frames count on the value of this pixel during the process. of decoding. According to a first variant, information representative of the use of a pixel for temporal prediction is a binary value indicating whether this pixel is used for temporal prediction (binary value set to one) or not (binary value set to zero). For example, for each pixel in the frame, information representative of the use of this pixel for temporal prediction is initialized to zero at the beginning of GOP decoding and is set to one each time this pixel is used for decoding a macroblock of a GOP. another picture. According to a second variant, information representative of the use of a pixel for temporal prediction characterizes the number of times this pixel is used for temporal prediction. For example, for each pixel in the frame, information representative of the use of this pixel for temporal prediction is initialized to zero at the beginning of GOP decoding and is incremented by one each time this pixel is used for decoding a macroblock of a GOP. another picture. In a third variant, information representative of the use of a pixel for temporal prediction is an accumulative prediction weight and thus carries more accurate information. For example, for each pixel in the frame, information representative of the use of this pixel for temporal prediction is initialized to zero at the beginning of GOP decoding and is incremented by the weight used during prediction. (Since a pixel can be predicted as a weighted average. two different pixels coming from two different reference frames) or during interpolation (since a pixel can be predicted by a subpixel coming from the interpolation of several reference pixels), each time this pixel is used for decoding of a macroblock from another frame. In other variants, the heatmap can also load added information! relating to macroblocks that use a pixel as a reference, for example, a list containing the frame number and predicted pixel position (which uses a particular reference pixel).

[OSOJAdvantageously. step S12 and step S14 are executed in parallel. Thus a single GOP decoding is performed to obtain spatial propagation maps and heat maps. However, the method is compatible with steps executed sequentially in any order, thus requiring separate decoding.

Figure 4 provides a detailed view of step S14 of the flowchart shown in Figure 3 according to a specific and non-limiting embodiment of the present principles. In addition, Figure 8 illustrates heat map determination according to a specific and non-limiting embodiment of the present principles. The figures are described together. A two-dimensional representation of frame i and the corresponding heat map HMi is used. According to the present representation, binary values are used in the heat map to represent the use of one pixel, or more precisely pixels of an MB macroblock, in a temporal prediction. An i frame is divided into the macroblock MB comprising N pixels. In Figure 6, black macroblock 600 corresponds to a macroblock 610 not yet used for temporal prediction, while white mawroblock 620 corresponds to a macroblock already for temporal prediction 630. However, the present principles are not limited to this convention and a The table comprising integer values, real values, multi-level color coding can be used. The process of step S14 is repeated during decoding of each image group. Within an image group, a frame counter is initialized as one and incremented by one until it reaches the end of the group. In a step S140, the current frame i is tested to determine if the associated heat map HMi needs to be created. For example, the GOP shown in FIG. 6 first comprises a reference table I, secondly a reference table P and thirdly a non-reference table B (indicated by b). As stated above, the heatmap is relevant only for reference frames, such as reference frames I, P and B, and, in a refinement, for GOP reference frames containing candidate elements. In a variant where candidate elements are motion vectors, heat map HM1 is not constructed for frame I (no candidate elements), as well as for heat map HMi associated with frame i = 3 (b frame no). of reference). In practice, if test S140 is positive (YES) for an input frame i, the corresponding heat map HMi is created and initialized in one step S142. Information representative of the use of a pixel for temporal prediction is set to zero for all pixels in the frame. If test S140 is negative (NO), step S142 is skipped. In one embodiment, this step S142 is not skipped and each heatmap value of a non-reference frame is set to a null value, for example -1 or 0, and subsequently ignored ie these heatmapps. are not updated during decoding of the following GOP frames. In step S144, the current frame is decoded, one macroblock MB after another. MB macrobios are sequentially decoded from the first upper left MB to the last lower right MB, taking Stew to record the dependence between macrobios by temporal prediction. As shown in Fig. 6, decoding the MB macroblock at frame i = 3 uses, for example, macroblock 630 from frame P (2) and macroblock 640 from frame I (i = 1). Thus, in one step 146, information representative of the use of a pixel for temporal prediction corresponding to the macroblock 630 pixels in heat map HM2 is set to one (white macroblock 620). In contrast, the macroblock in the heat map HM1 is left unchanged. In fact, in a variant where candidate elements are motion vectors, no time propagation information is required for frame I pixels since they do not host candidate elements. As a result, in this step 146, the heat maps associated with the previously decoded frames, between 1 and s-1, are updated sequentially according to the dependence of the decoded MB on the current frame i = 3. The process is repeated for each successive GOP frame. While decoding each GOP frame, each relevant heatmap is updated, which comprises at least one and potentially up to 12 H.264 heatmaps as up to 12 reference frames are defined in the pattern.

Returning to Figure 3, in step S16, modifiable elements between candidate modifiable elements of the encoded bit stream are selected based on the spatial propagation map of the modifiable element and the corresponding heat map. Figure 5 details this flowchart step S16 according to a specific and non-limiting embodiment of the present principles. Moreover, Figure 7 illustrates the selection of modifiable elements according to a specific and non-limiting embodiment of the present principles. Figures 5 and 7 are described together. Step S16 is performed by processing the GOP in reverse order from the last frame i = n to the first frame i = 1. Step S16 comprises mainly two substeps. First, candidate modifiable elements that are compatible with propagation-uspacial (propagation map) and temporal (heat map) compatibility are selected as watermarks, and secondly, heat maps are updated according to the referenced pixels. by the macrobtocks of the spatial propagation map associated with the alternative value. In short, for non-reference frames (type b), the selection strategy is one used in the classic watermarked signaling system as described in WO 2010/021882 A1. Those skilled in the art will appreciate that, for each candidate position, the most invisible and robust candidate modifiable elements, whose spatial propagation does not overlap with candidate modifiable elements previously placed on the candidate list, are preferably selected as a watermark. Thus, in the example GOP of FIG. 7, the last frame i = 6 of the GOP is a frame P not used as a reference since it is the last frame of the GOP, and each modifiable macroblock 700 which is modified according to an alternative value of a motion vector will not propagate temporally to other frames. The selection strategy described above is applied.

[Q33] In contrast, for reference frames (type P and B), the selection strategy is applied with another constraint that considers temporal propagation (as processed in step S14). For example, in the example GOP of Figure 7, GOP frame i = 4 is a P frame used as a reference. Steps S160, S162 and S164 are repeated sequentially for each modifiable macroblock 710, 720, 730 associated with a candidate modifiable element. Thus, in a preliminary step, a modifiable macroblock 710 of a current frame i is obtained for evaluation. In a first step S160, when the spatial propagation map of the candidate modifiable element (associated with the modifiable macroblock 710 or 730) overlaps with the heat map HM4 associated with the processed frame, the modifiable element (associated with the modifiable macroblock 710 or 730) is not selected as a possible watermark. This is represented in Figure 7 by a scratched block 710 or 730. In the binary representation of Figure 7, "overlap" means that the position of at least one modified pixel in the spatial propagation map because of the modified element is identical to the position. In other words, candidate modifiable elements are discarded when the heatmap value of at least one pixel in the spatial spread map, virtually mapped in the heatmap (715 or 735), is not By contrast, when the spatial propagation map of the candidate modifiable element (associated with modifiable macroblock 720) does not overlap with heatmap 725 in the processed frame HM4, the modifiable element (associated with modifiable macroblock 720) ) is selected in one step S162 as a possible watermark, so in table 1 = 4 two modifiable elements are selected. , this variant where the selected watermarks are composed only of unused pixels (pixels with 0 values in black in the corresponding heat map as shown in Figure 7) avoids any temporal propagation by construction. (034] However, in other variants, selection may be more flexible and may allow limited time propagation by adding candidate modifiable elements to the selection having some or all pixels with a low reference score, that is, a low value on the map. For example, if a candidate modifiable element uses a pixel with a score of 1 or less (the other being 0), it will propagate temporally to another frame / pixel that may be known if the heatmap carries information. Therefore, in a first variant, for each candidate modifiable element, a first dependency value comprising the number of pixels in the spatial propagation map of the current candidate modifiable element that are used for temporal prediction according to the heat map is computed, then all candidate modifiable elements having this first value d and dependence greater than a first threshold are discarded. In this embodiment, the candidate modifiable element (resulting in the modifiable macroblock 730) having only a small overlap with the upper map is selected. According to a second variant compatible with non-binary upper maps, for each candidate modifiable element, a second dependency value comprising the sum of information representative of the use of a temporal prediction pixel associated with the pixels in the spatial propagation map of the element. modifiable candidate is computed. Subsequently, all candidate modifiable elements having this second dependency value greater than a second threshold are discarded. According to this variant, a candidate modifiable element is selected as long as it impacts only a small number of macroblocks.

[035] As a result, the step adds! S18 allows, depending on the described variant, to prevent or control the temporal propagation of the inserted watermarks.

In addition, in a final step S164, information representative of the use of a pixel for temporal prediction is updated for all pixels used as a macroblock reference of the spatial propagation map associated with the modified value of the selected candidate modifiable element. In fact, since the watermarked signaling system modifies predictions, for example by modifying the motion vector, it also modifies the way in which reference frames are used, and thus heat maps. of reference, but in a small way. For example, a macroblock 720 of the spatial propagation map associated with a selected modified value is shown in figure 7 in table i = 4. Depending on whether the original or alternate motion vector is used, the temporal prediction has one of two macroblocks in the lower left of frame i = 2 and the heatmap therefore needs to be update 745 to ensure that both macroblocks are preserved. More precisely, the port 745 associated with the original motion vector appears in heat map HM2 at step S146 while the other motion vector appears during update step S164 once the corresponding modifiable element has been selected. Thus, for each selected watermark, heat maps associated with frames referenced by the spatial propagation map macroblock need to be updated to consider this potential alternative configuration.

[037] The implementations described herein may be implemented, for example, in a method or process, an apparatus, a software program, a data stream or a signal. Even if discussed only in the context of a single form of implementation (for example, discussed only as a method or a device), the feature implementation discussed can also be implemented in other forms (for example, a program). A device may be implemented, for example, in appropriate hardware, software and firmware. The methods may be implemented, for example, in an apparatus such as, for example, a processor, which refers to processing devices in general, including, for example, a computer, a microprocessor, an integrated circuit or a processor. programmable logic device. Processors also include communication devices such as computers, mobile phones, portable / personal digital assistants ("PDAs"), and other devices that facilitate information communication between end users.

Implementations of the various processes and features described herein may be incorporated into a variety of different equipment or applications, particularly, for example, equipment or applications. Examples of such equipment include an encoder, a decoder, a postprocessor processing output from a decoder, a preprocessor providing input to an encoder, a video encoder, a video decoder, a video codec, a network server. , a signal converter apparatus, a laptop, a personal computer, a cell phone, a PDA, and other communication devices. OK! As should be clear, the equipment can be mobile and even installed in a mobile vehicle.

Additionally, the methods may be implemented by instructions being executed by a processor, and such instructions (and / or data values produced by an implementation} may be stored on processor-readable media such as, for example, an integrated circuit, software carrier, or other storage device such as, for example, a hard disk, compact disc ("CD"), optical disc (such as, for example, a Blu-ray, often DVD digital versatile disk or digital video disc), random access memory (“RAM”), or read-only memory (“ROM”). The instructions may form a tangibly embedded application program in readable media per processor Instructions can be for example in hardware, firmware, software or in a combination Instructions can be found in for example an operating system, a separate application or Thus, a processor may be characterized, for example, as either a device configured to perform a process or a device that includes processor readable media (such as a storage device) having instructions for performing a process. process. Additionally, processor-readable media can store, in addition to or instead of instructions, data values produced by an implementation. -404031a! As will be apparent to those skilled in the art, implementations may produce a variety of signals formatted to carry information that may be, for example, stored or transmitted. The information may include, for example, instructions for executing a method, or data produced by one of the described implementations. For example, a signal may be formatted to load as rules the rules for writing or reading the syntax of a described mode, or for loading as data the actual syntax values recorded by a described mode. Such a signal may be formatted, for example, as an electromagnetic wave (for example, using a radio frequency part of the spectrum) or as a baseband signal. Formatting may include, for example, encoding a data stream and modulating a carrier with the encoded data stream. The information that the signal carries may be, for example, analog or digital information. The signal may be transmitted over a variety of different wired or wireless connections as is known. The signal can be stored on processor readable media.

[041] Several implementations have been described. Nevertheless, it will be understood that various modifications may be made. For example, elements of different implementations may be combined, supplemented, modified or removed to produce other implementations. Additionally, one of ordinary skill in the art will understand that other structures and processes may be substitutes for those disclosed and the resulting implementations will perform at least substantially the same function (s), at least substantially the same (s). ) method (s) to achieve at least substantially the same result (s) as the disclosed implementations. Therefore, these and other implementations are considered by this request. nit CLAIMS

Claims (14)

Method for determining a set of modifiable elements of an encoded bit stream representative of audio and video content, the method characterized by the fact that it comprises, for a portion of said encoded bit stream encoding a group of images: determining (S12) a set of candidate modifiable elements, wherein a candidate modifiable element comprises a modified value for a bit stream segment encoded at a given location and a spatial propagation map associated with said modified value, said propagation map spatial (SP6) comprising pixels whose decoding is Impacted when said modified value is placed in the bit stream encoded in said given location; determining (S14) a heat map (HMi) for each reference frame comprising, for each pixel of said reference frame, information representative of the use of said pixel for temporal prediction during decoding of said portion of said encoded bit stream. encoding a group of images; obtaining (S16) a set of modifiable elements between said set of candidate modifiable elements, wherein a modifiable element has a spatial propagation map that does not overlap the corresponding heat map. Method according to claim 1, characterized in that determining said information representative of the use of said pixel for temporal prediction comprises a binary value indicating whether the pixel is used for said temporal prediction. Method according to claim 1, characterized in that determining said information representative of the use of said pixel for temporal prediction comprises the number of times the pixel is used for said temporal prediction. Method according to claim 1, characterized in that determining said information representative of the use of said pixel for temporal prediction comprises a cumulative weight of the pixel used for prediction. Method according to claim 1, characterized in that said reference frames comprise P frames and B reference frames. Method according to any one of claims 1 to 5, characterized in that determining a heat map comprises for each frame of said group of images: initializing (S142) said information representative of the use of a pixel to temporal prediction; decoding (S144) said frame; and for each decoded macrobocus of said decoded frame, updating (S146) said information representative of the use of a pixel for temporal prediction when said pixel is used as a reference for said decoded macroblock. A method according to claim 8, characterized in that obtaining a modifiable element comprises, for each frame of said group of scanned images in reverse order: selecting (S160, S162) a candidate modifiable element whose spatial propagation map does not override the heat map of said frame; updating (S164) said information representative of the use of a pixel for temporal prediction for all macroblock pixels of said spatial propagation map associated with said modified value of said selected candidate modifiable element. Method according to claim 7; CHARACTERIZED by the fact that selecting a candidate modifiable element (i includes discarding candidate modifiable elements when the heatmap value of at least one pixel in the sword propagation map) is not equal to zero. A method according to claim 7, characterized in that selecting a candidate modifiable element comprises: for each candidate modifiable element, determining a first dependency value comprising the number of pixels in the spatial propagation map of said candidate modifiable element. which are used for prediction according to heat map; discard all candidate modifiable elements having said first dependency value greater than a first threshold. A method according to claim 7 or claim 9, characterized in that selecting a candidate modifiable element comprises: for each candidate modifiable element, determining a second dependency value comprising the sum of information representative of the use of a pixel for temporal prediction associated with the pixels in the spatial propagation map of said candidate modifiable element; discard all candidate modifiable elements having said second dependency value greater than a second threshold, An apparatus for determining a set of modifiable elements of an encoded bit stream representative of audio and video content, characterized in that it comprises at least one processor (110) configured for a portion of the encoded bit stream encoding a group of images for: determining a set of candidate modifiable elements, wherein a candidate modifiable element comprises a modified value for a bit stream segment encoded at a given location and a spatial propagation map associated with said modified value, said spatial propagation map (SP6) comprising pixels whose decoding is impacted when said modified value is placed in the bit stream encoded in said given location; determining a heat map for each reference frame comprising, for each pixel of said reference frame, information representative of the use of said pixel for temporal prediction during decoding said portion of said encoded bit stream encoding a group of images; obtain a set of modifiable elements between said set of candidate modifiable elements in which a modifiable element has a spatial propagation map that does not overlap the corresponding heat map. A device for determining a set of modifiable elements of an encoded bit stream representative of audio and video content, the device characterized by that it comprises for a portion of the encoded bit stream encoding a group of images; resources for determining (12) a set of candidate modifiable elements, wherein a candidate modifiable element comprises a modified value for a bit stream segment encoded at a given location and a spatial propagation map associated with said modified value, said spatial propagation map (SP6) comprising pixels whose decoding is impacted when said modified value is placed in the bit stream encoded in said given location; resources for determining (14) a heat map for each reference frame, said heat map comprising, for each pixel of a reference frame, information representative of the use of said pixel for temporal prediction during the decoding of said part of the frame. said encoded bit stream encoding a group of images; and resources for obtaining (16) a set of modifiable elements between said set of candidate modifiable elements, wherein a modifiable element has a spatial spread map that does not overlap the corresponding heat map. Computer program product, characterized in that it comprises program code instructions for performing the method steps according to any one of claims 1 to 10 when this program is executed on a computer. Processor readable media, characterized in that it has instructions stored therein to make a processor perform at least the method steps according to any one of claims 1 to 10.